Load Balancing Microsoft Exchange Deployment Guide v Copyright Loadbalancer.org, Inc

Transcription

1 Load Balancing Microsoft Exchange 2013 Deployment Guide v1.4.3 Copyright Loadbalancer.org, Inc

2 Table of Contents 1. About this Guide Loadbalancer.org Appliances Supported Loadbalancer.org Software Versions Supported Microsoft Exchange Software Versions Supported Exchange Server Exchange 2013 Server Roles Load Balancing Exchange Load Balancing & HA Requirements... 6 Client Access Array... 6 Database Availability Group (DAG)... 6 Persistence (aka Server Affinity)... 6 Port Requirements... 6 SSL Termination... 7 HTTPS Namespaces & IP addresses... 7 Health-Checks Deployment Concept...8 Virtual Service (VIP) Requirements Load Balancer Deployment Methods...9 Layer 4 DR Mode... 9 Layer 7 SNAT Mode Our Recommendation Configuring Exchange 2013 for Load Balancing ) External Access Domain ) Virtual Directories ) Outlook Anywhere ) Autodiscover ) Certificates ) Send & Receive Connectors Adding Connectors ) DNS Configuration ) Additional Configuration Steps (depends on Load balancing method)...16 DR Mode NAT Mode SNAT Mode ) IIS Restart (** Important **) Loadbalancer.org Appliance the Basics...17 Virtual Appliance Download & Deployment Initial Network Configuration Accessing the Web User Interface (WebUI) HA Clustered Pair Configuration Appliance Configuration for Exchange 2013 Using DR Mode...20 Configure VIP1 CAS Role HTTPS Services Configure HTTP to HTTPS OWA Redirect... 21

3 Configure VIP2 CAS Role IMAP4/POP3 Services...21 Configure VIP3 CAS Role SMTP Services Appliance Configuration for Exchange 2013 Using SNAT Mode...24 Configure Layer 7 Timeout Settings Configure VIP1 CAS Role HTTPS Services Configure HTTP to HTTPS OWA Redirect Configure VIP2 CAS Role IMAP4/POP3 Services...26 Configure VIP3 CAS Role SMTP Services Finalizing the Configuration Testing & Verification...30 Useful Exchange 2013 & Other Microsoft Tools Testing Server Health-checks using Set-ServerComponentState...30 Testing Mailflow Testing SMTP Mail flow using Telnet Microsoft Exchange Testing Tool Useful Appliance based Tools & Features Using System Overview Layer 4 Status Report Layer 7 Statistics Report Appliance Logs Technical Support Further Documentation Conclusion Appendix Enabling Layer 7 Transparency using TProxy Limiting inbound SMTP Connections using Firewall Rules Using a Layer 4 Virtual Service for SMTP Configuring an HTTP to HTTPS redirect for OWA Clustered Pair Configuration Adding a Slave Unit Solving the ARP Problem Company Contact Information... 44

4 About this Guide 1. About this Guide This guide details the steps required to configure a load balanced Microsoft Exchange 2013 environment utilizing Loadbalancer.org appliances. It covers the configuration of the load balancers and also any Microsoft Exchange 2013 configuration changes that are required to enable load balancing. For more information about initial appliance deployment, network configuration and using the Web User Interface (WebUI), please also refer to the relevant Administration Manual: v7 Administration Manual v8 Administration Manual 2. Loadbalancer.org Appliances Supported All our products can be used with Exchange The complete list of models is shown below: Discontinued Models Current Models * Enterprise R16 Enterprise R20 Enterprise VA R16 Enterprise MAX Enterprise VA Enterprise 10G Enterprise R320 Enterprise Ultra Enterprise VA R20 Enterprise VA MAX Enterprise AWS ** Enterprise AZURE ** * For full specifications of these models please refer to: ** Some features may not be supported, please check with Loadbalancer.org support 3. Loadbalancer.org Software Versions Supported V7.6.4 and later 4. Microsoft Exchange Software Versions Supported Microsoft Exchange 2013 CU2 ( ) and later Page 4

5 Exchange Server Exchange Server 2013 Exchange 2013 is Microsoft's latest enterprise level messaging and collaboration server. Exchange 2013 has been designed for simplicity of scale, hardware utilization, and failure isolation. This has greatly simplified both the deployment process and the implementation of a load balancer. 6. Exchange 2013 Server Roles Exchange 2013 has been consolidated into two roles, these are: the Client Access Server role and the Mailbox Server role. The functionality of the Hub Transport server role has split between the CAS role (the Front End Transport Service) and the Mailbox Server role (the Transport Service and the Mailbox Transport Service). The Edge Transport server role has been removed. Role Purpose Client Access Server The Client Access Server role is comprised of three components, client protocols, SMTP, and a UM Call Router. The CAS role is a thin, protocol session stateless server that is organized into a load balanced configuration. Unlike previous versions, session affinity is not required at the load balancer. This is because logic now exists in CAS to authenticate the request, and then route the request to the Mailbox server that hosts the active copy of the mailbox database. A number of issues have been seen with IOS-7 on the iphone when used with ActiveSync. Upgrading to IOS-8 resolved these issues. Mailbox Server The Mailbox Server role now hosts all the components and/or protocols that process, render and store the data. No clients will ever connect directly to the Mailbox server role; all client connections are handled by the Client Access Server role. Mailbox Servers can be added to a Database Availability Group, thereby forming a high available unit that can be deployed in one or more datacenters. CAS Array Object This concept has been removed and there is no longer any need to define a CAS array object. Client Access Protocols Outlook clients no longer use RPC to access their mailbox. This is now handled only by RPC over HTTPS (aka Outlook Anywhere). Native RPC is only used for server to sever communication. POP3 and IMAP4 continue to be supported as with previous versions. External SMTP Mail flow External SMTP communication is now handled by the Front End Transport Service on the CAS role. Exchange Administration The Exchange Admin Center (EAC) is the new web-based management console in Microsoft Exchange Server The EAC replaces the Exchange Management Console (EMC) and the Exchange Control Panel (ECP), which were the two interfaces used to manage Exchange Server Note that ECP is still the name of the virtual directory used by the EAC. Page 5

6 Load Balancing Exchange Load Balancing Exchange 2013 It's highly recommended that you have a working Exchange 2013 environment first before implementing the load balancer. LOAD BALANCING & HA REQUIREMENTS In Exchange Server 2013, there are two basic building blocks the Client Access Array and the Database Availability Group (DAG). Each provides a unit of high availability and fault tolerance that are decoupled from one another. Multiple Client Access Servers make up the Client Access Array, while multiple Mailbox Servers form the DAG. CLIENT ACCESS ARRAY As mentioned earlier, the 2010 concept of a CAS Array no longer exists. In 2013, a Client Access Array is simply a group of two or more Client Access Servers. The load balancer then enables resilience and HA. DATABASE AVAILABILITY GROUP (DAG) A DAG is a group of up to 16 Mailbox Servers that hosts a set of databases and provides automatic database-level recovery from failures that affect individual servers or databases. DAG's utilize Microsoft Clustering Services which cannot be enabled on the same server as Microsoft Network Load Balancing (NLB). Therefore, using Microsoft NLB is not an option in this case. Using a Loadbalancer.org hardware or virtual appliance provides an ideal solution. PERSISTENCE (AKA SERVER AFFINITY) Due to Exchange 2013's new architecture, all sessions to the CAS servers are stateless and therefore persistence/affinity is no longer required on the load balancer. PORT REQUIREMENTS The following table shows the port list that must be load balanced. Some services such as IMAP4 or POP3 may not be used in your environment. TCP Port Role(s) Uses 25 CAS Inbound SMTP 110 CAS POP3 clients 143 CAS IMAP4 clients 443 CAS HTTPS (Outlook Web App, AutoDiscovery, Web Services, ActiveSync, Outlook Anywhere, Offline Address Book, Exchange Administration Center) 993 CAS Secure IMAP4 clients 995 CAS Secure POP3 clients Page 6

7 Load Balancing Exchange 2013 SSL TERMINATION SSL offloading for Exchange 2013 is supported from SP1 as detailed in this Microsoft article. However, for scalability and effective load sharing we recommend terminating SSL on the Exchange Servers rather than on the load balancer. HTTPS NAMESPACES & IP ADDRESSES The following examples show 2 different approaches to HTTPS namespace configuration and the related load balancing considerations for each. Example 1 simple namespace configuration Namespace Purpose mail.robstest.com Outlook Web App, ActiveSync, Outlook Anywhere, Offline Address Book, Exchange Web Services autodiscover.robstest.com Auto Discover Notes: In this case a single VIP is used for all HTTPS namespaces/services Both DNS entries should then point at the same VIP This method is simple to setup, but only permits a single Exchange URL to be health checked. However, a successful full HTTPS service check on the OWA virtual directory is a good indication that the other Virtual Directories & applications are also functioning correctly Example 2 expanded namespace configuration Namespace Purpose owa.robstest.com Outlook Web Access outlook.robstest.com Outlook Anywhere ews.robstest.com Exchange Web Services autodiscover.robstest.com Autodiscover activesync.robstest.com ActiveSync oab.robstest.com Offline Address Book Notes: In this case multiple VIPs are used one for each HTTPS namespace/service This guide uses the config of example 1 above, i.e. a single IP address for all services. Each related DNS entry should then point at the corresponding VIP This method is more complex to setup, but does enable more granular health checks to be configured Page 7

8 Load Balancing Exchange 2013 HEALTH-CHECKS In this guide, the health check for HTTPS services accesses owa/healthcheck.htm on each server and checks for a '200 OK' response. A different virtual directory (e.g. ECP, EWS etc.) can be chosen if preferred or more appropriate. Note that healthcheck.htm is generated in-memory based on the component state of the protocol in question and does not physically exist on disk. 8. Deployment Concept There are multiple ways to deploy Exchange, but in this example two servers are used. Each server hosts the CAS & Mailbox roles in a DAG configuration. This provides high availability and uses a minimum number of Exchange Servers. Clients then connect to the Virtual Services (VIPs) on the load balancer rather than connecting directly to one of the Exchange servers. These connections are then load balanced across the Exchange servers to distribute the load according to the load balancing algorithm selected. DAG DAG client connections CAS/MBOX 1 VIPs LB Clustered Pair CAS/MBOX 2 VIPs = Virtual IP Addresses The load balancer can be deployed as a single unit, although Loadbalancer.org recommends a clustered pair for resilience & high availability. Please refer to section 6 in the appendix on page 38 for more details on configuring a clustered pair. VIRTUAL SERVICE (VIP) REQUIREMENTS To provide load balancing and HA for Exchange 2013, the following VIPs are required: HTTPS (for all HTTPS based services) SMTP Optionally, additional VIPs may be required as follows: HTTP (for redirecting to HTTPS, see page 37 in the appendix for more details) IMAP4 POP3 Page 8

9 Deployment Concept IMAP4 and POP3 are not typically used. Therefore these VIPs are not generally required. 9. Load Balancer Deployment Methods The load balancer can be deployed in one of 3 fundamental ways; Layer 4 DR mode, Layer 4 NAT mode or Layer 7 SNAT mode. For Exchange, layer 4 DR mode or layer 7 SNAT mode is recommended; these are described below. Layer 4 NAT mode is rarely used for Exchange, please refer to the Administration Manual for details of this mode. LAYER 4 DR MODE One-arm direct routing (DR) mode is a very high performance solution that requires little change to your existing infrastructure. Kemp, Brocade, Barracuda & A10 Networks call this Direct Server Return and F5 call it N-Path. DR mode works by changing the destination MAC address of the incoming packet to match the selected Real Server on the fly which is very fast When the packet reaches the Real Server it expects the Real Server to own the Virtual Services IP address (VIP). This means that you need to ensure that the Real Server (and the load balanced application) respond to both the Real Servers own IP address and the VIP The Real Server should not respond to ARP requests for the VIP. Only the load balancer should do this. Configuring the Real Servers in this way is referred to as Solving the ARP Problem. Please refer to page 40 for more information On average, DR mode is 8 times quicker than NAT for HTTP, 50 times quicker for Terminal Services and much, much faster for streaming media or FTP The load balancer must have an Interface in the same subnet as the Real Servers to ensure layer 2 Page 9

10 Load Balancer Deployment Methods connectivity required for DR mode to work The VIP can be brought up on the same subnet as the Real Servers, or on a different subnet provided that the load balancer has an interface in that subnet Port translation is not possible in DR mode i.e. having a different RIP port than the VIP port DR mode is transparent, i.e. the Real Server will see the source IP address of the client LAYER 7 SNAT MODE Layer 7 SNAT mode uses a proxy (HAProxy) at the application layer. Inbound requests are terminated on the load balancer, and HAProxy generates a new request to the chosen Real Server. As a result, Layer 7 is a slower technique than DR or NAT mode at Layer 4. Layer 7 is typically chosen when either enhanced options such as SSL termination, cookie based persistence, URL rewriting, header insertion/deletion etc. are required, or when the network topology prohibits the use of the layer 4 methods. This mode can be deployed in a one-arm or two-arm configuration and does not require any changes to the Real Servers. However, since the load balancer is acting as a full proxy it doesn't have the same raw throughput as the layer 4 methods. The load balancer proxies the application traffic to the servers so that the source of all traffic becomes the load balancer. SNAT mode is a full proxy and therefore load balanced Real Servers do not need to be changed in any way Because SNAT mode is a full proxy any server in the cluster can be on any accessible subnet including across the Internet or WAN SNAT mode is not transparent by default, i.e. the Real Servers will not see the source IP address of the client, they will see the load balancers own IP address by default, or any other local appliance IP address if preferred (e.g. the VIP address), this can be configured per layer 7 VIP. If required, the clients IP address can be passed through either by enabling TProxy on the load balancer, or for HTTP, using X-forwarded-For headers. Please refer to chapter 6 in the Administration Manual for Page 10

11 Load Balancer Deployment Methods more details SNAT mode can be deployed using either a 1-arm or 2-arm configuration OUR RECOMMENDATION Where possible we recommend that Layer 4 Direct Routing (DR) mode is used. This mode offers the best possible performance since replies go directly from the Real Servers to the client, not via the load balancer. It's also relatively simple to implement. Ultimately, the final choice does depend on your specific requirements and infrastructure. If DR mode cannot be used, for example if the Exchange servers are located in remote routed networks then SNAT mode is recommended. It's important to remember that when using SNAT mode (HAProxy), the source IP address of packets reaching the Exchange Servers will be the IP address of the load balancer and not the source IP address of the client. Transparency is normally only an issue for SMTP traffic at the receive connector. System Administrators typically want to lock down receive connectors to accept SMTP connections only from a controlled set of devices such as external smart mail hosts, printers, networked photocopiers etc. If transparency for SMTP is the only issue, there are a number of options available to address this: Option 1 Enable full layer 7 transparency using TProxy. This is covered in section 1 of the Appendix on page 35. Option 2 Use the load balancers on-board firewall to lock down inbound SMTP connections rather than doing this at the receive connector. This is covered in section 2 of the Appendix on page 36. Option 3 Configure a layer 4 Virtual Service for SMTP rather than a layer 7 (HAProxy) based Virtual Service. Layer 4 is transparent by default so the source IP address is maintained. This is covered in section 3 of the Appendix on page 36. Page 11

12 Configuring Exchange 2013 for Load Balancing 10. Configuring Exchange 2013 for Load Balancing 1) EXTERNAL ACCESS DOMAIN This can be configured using the EAC. Select servers > virtual directories and then click the spanner icon. This will open the form shown below. All CAS servers should be configured with a valid external name, e.g. mail.robstest.com 2) VIRTUAL DIRECTORIES The Internal and External URL's for the various virtual directories need to be configured to suit your environment. The External URL's are automatically set to be the same as the external access domain when this is configured, but can be changed if needed. The Internal URL's must be set individually by clicking the Edit (pen) icon for each virtual directory. All settings can be configured using the EAC option: servers > virtual directories as shown below: Page 12

13 Configuring Exchange 2013 for Load Balancing 3) OUTLOOK ANYWHERE This is configured using the EAC. Select servers > servers and then click the edit (pen) icon next to each sever, click the Outlook Anywhere option as shown below to change the setting. The external and internal names for each server should be configured as required, e.g. mail.robstest.com 4) AUTODISCOVER Internal A new Active Directory object named the service connection point (SCP) is created on the server where you install the Client Access Server role. The SCP object contains the authoritative list of Autodiscover service URLs for the forest. The Set-ClientAccessServer cmdlet is used to update the SCP object as shown in the following example: Set-ClientAccessServer -Identity "EXCH01" -AutoDiscoverServiceInternalUri " Once configured, the Test AutoConfiguration option available when <CTRL> right-clicking the Outlook icon in the taskbar can be used to view these settings as shown below: The minimum Outlook client for Exchange 2013 is Outlook 2007, 2003 is NOT supported. Page 13

14 Configuring Exchange 2013 for Load Balancing External When Outlook is started on a client that is not domain-connected, it first tries to locate the Autodiscover service by looking up the SCP object in Active Directory. Because the client is unable to contact Active Directory, it tries to locate the Autodiscover service by using DNS. In this scenario, the client will determine the domain of the user s address, and then check DNS by using two predefined URLs. For the SMTP domain robstest.com, Outlook will try the following two URLs to try to connect to the Autodiscover service: Again, this can be seen using the Test AutoConfiguration option as shown below: Page 14

15 Configuring Exchange 2013 for Load Balancing 5) CERTIFICATES The recommended approach is to use SAN certificates and specify all required namespaces. It's also possible to use wildcard certs if preferred. Certificate requests can be generated using either the graphical based Exchange Admin Center or the command based Exchange Management Shell. The EAC can also be used to import/export certificates using the server > certificates > More option ** IMPORTANT!! - the same certificate and private key must be deployed on all Exchange Servers ** SSL offloading for Exchange 2013 is supported from SP1 as detailed in this Microsoft article. However, for scalability and effective load sharing we recommend terminating SSL on the Exchange Servers rather than on the load balancer. 6) SEND & RECEIVE CONNECTORS By default no send connectors are created when Exchange 2013 is installed. A send connector must be created manually that either sends outbound messages to a smart host or directly to their recipient using DNS. For a dual role server that has both the CAS and Mailbox roles, five receive connectors are automatically created by default. The table below lists these connectors: Receive Connector Role Purpose Default <server name> Mailbox Accepts connections from Mailbox servers running the Transport service and from Edge servers Client Proxy <server name> Mailbox Accepts connections from front-end servers. Typically, messages are sent to a front-end server over SMTP Default FrontEnd <server name> CAS Accepts connections from SMTP senders over port 25. This is the common messaging entry point into your organization Outbound Proxy Frontend <server name> CAS Accepts messages from a Send Connector on a back-end server, with front-end proxy enabled Client Frontend <server name> CAS Accepts secure connections, with Transport Layer Security (TLS) applied For more information on mail connectors please refer to the following Technet article: ADDING CONNECTORS Connectors can be created using the Exchange Administration Center (EAC) or the Exchange Management Shell. Receive connectors must use a unique combination of IP address bindings, port number assignments, and remote IP address ranges from which mail is accepted. Multiple send connectors can created, this is typically done to enables multiple outbound routes to specified that have different costs. The exact connector configuration depends on your specific environment and requirements. Page 15

16 Configuring Exchange 2013 for Load Balancing 7) DNS CONFIGURATION Configure appropriate internal and external DNS entries for the various Internal and External URL's that have been defined in steps 1) to 4). The DNS entries should point at the HTTPS VIP on the load balancer assuming a simple namespace design as shown below: DNS record Purpose mail.robstest.com Points at the VIP used for all HTTPS based services autodiscover.robstest.com Points at the VIP used for all HTTPS based services If multiple VIPs are defined for the various Virtual Directories, DNS should be configured accordingly. 8) ADDITIONAL CONFIGURATION STEPS (DEPENDS ON LOAD BALANCING METHOD) The steps required depend on the load balancing mode used as described below. DR MODE The 'ARP problem' must be solved on each Exchange Server for DR mode to work. For detailed steps on solving the ARP problem for the various versions of Windows, Please refer to page 40 for more information. NAT MODE When using Layer 4 NAT mode, the default gateway on each Exchange Server MUST be set to be the loadbalancer. It's recommended that a floating IP address is used rather than the interface IP address. This makes it possible for the load balancer to failover to a slave unit and successfully bring up the gateway address. SNAT MODE When using SNAT mode, no configuration changes to the Exchange Servers are required. 9) IIS RESTART (** IMPORTANT **) Once all Exchange configuration is complete restart IIS on each server (or reboot the server) to ensure all changes are applied. This can be done using the following command in a command or Powershell Window: iisreset /restart Page 16

17 Loadbalancer.org Appliance the Basics 11. Loadbalancer.org Appliance the Basics VIRTUAL APPLIANCE DOWNLOAD & DEPLOYMENT A fully featured, fully supported 30 day trial is available if you are conducting a PoC (Proof of Concept) deployment. The VA is currently available for VMware, Virtual Box, Hyper-V, KVM and XEN and has been optimized for each Hypervisor. By default, the VA is allocated 1 CPU, 2GB of RAM and has an 8GB virtual disk. The Virtual Appliance can be downloaded here. The same download is used for the licensed product, the only difference is that a license key file (supplied by our sales team when the product is purchased) must be applied using the appliance's WebUI. Please refer to the Administration Manual and the ReadMe.txt text file included in the VA download for more detailed information on deploying the VA using various Hypervisors. INITIAL NETWORK CONFIGURATION The IP address, subnet mask, default gateway and DNS settings can be configured in several ways as detailed below: Method 1 - Using the Network Setup Wizard at the console After boot up, follow the instructions on the console to configure the IP address, subnet mask, default gateway and DNS settings. Method 2 - Using the WebUI Using a browser, connect to the WebUI on the default IP address/port: To set the IP address & subnet mask, use: Local Configuration > Network Interface Configuration To set the default gateway, use: Local Configuration > Routing To configure DNS settings, use: Local Configuration > Hostname & DNS Method 3 - Using Linux commands At the console, set the initial IP address using the following command: ip addr add <IP address>/<mask> dev eth0 At the console, set the initial default gateway using the following command: route add default gw <IP address> <interface> At the console, set the DNS server using the following command: echo nameserver <IP address> >> /etc/resolv.conf If method 3 is used, you must also configure these settings using the WebUI, otherwise the settings will be lost after a reboot. Page 17

18 Loadbalancer.org Appliance the Basics ACCESSING THE WEB USER INTERFACE (WEBUI) The WebUI can be accessed via HTTP at the following URL: * Note the port number 9080 The WebUI can be accessed via HTTPS at the following URL: * Note the port number 9443 (replace with the IP address of your load balancer if it's been changed from the default) Login using the following credentials: Username: loadbalancer Password: loadbalancer To change the password, use the WebUI menu option: Maintenance > Passwords. Once logged in, the WebUI will be displayed as shown on the following page: Page 18

19 Loadbalancer.org Appliance the Basics (shows v8.2.x) HA CLUSTERED PAIR CONFIGURATION Loadbalancer.org recommend that load balancer appliances are deployed in pairs for high availability. In this guide a single unit is deployed first, adding a secondary slave unit is covered in section 6 of the Appendix on page 38. Page 19

20 Appliance Configuration for Exchange 2013 Using DR Mode 12. Appliance Configuration for Exchange 2013 Using DR Mode CONFIGURE VIP1 CAS ROLE HTTPS SERVICES a) Setting up the Virtual Service 2. Using the WebUI, navigate to: Cluster Configuration > Layer 4 Virtual Service and click Add a New Virtual Service Enter the following details: Enter an appropriate label for the VIP, e.g. CAS-HTTPS Set the Virtual Service IP address field to the required IP address, e.g Set the Virtual Service Ports field to 443 Leave Protocol set to TCP Leave Forwarding Method set to Direct Routing Click Update Now click Modify next to the newly created VIP Set Balance mode to Weighted Round Robin 1. Microsoft recommends that 'Round Robin' rather than 'Least Connection' should be used to help prevent over loading servers when they are brought online. This could occur if Least Connection was selected, since the load balancer would try to balance the number of connections across all real severs and therefore send all new requests to the new server. The trade off here is that using Round Robin will mean that server load may remain unbalanced for some time after bringing a new server into the active pool Un-check the Persistence option Set Check Type to Negotiate Set Protocol to HTTPS Set Request to send to owa/healthcheck.htm Page 20

21 Appliance Configuration for Exchange 2013 Using DR Mode As mentioned earlier, any other Exchange virtual directory (e.g. ECP, EWS etc.) can be used if preferred or more appropriate. All have an associated healthcheck.htm that can be used in the same way. Note that healthcheck.htm is generated in-memory based on the component state of the protocol in question and does not physically exist on disk. 15. Set Response expected to 200 OK 16. Click Update b) Setting up the Real Servers 2. Using the WebUI, navigate to: Cluster Configuration > Layer 4 Real Servers and click Add a new Real Server next to the newly created VIP Enter the following details: Enter an appropriate label for the RIP, e.g. CAS1 Change the Real Server IP Address field to the required IP address, e.g Click Update Repeat the above steps to add your other CAS Server(s) 1. CONFIGURE HTTP TO HTTPS OWA REDIRECT If required, the load balancer can be configured to automatically redirect users who attempt to connect to to For details on configuring this, please refer to section 4 in the Appendix on page 37. CONFIGURE VIP2 CAS ROLE IMAP4/POP3 SERVICES a) Setting up the Virtual Service These steps show IMAP4 settings, for POP3 change the port numbers from 143 & 993 to 110 & Using the WebUI, navigate to: Cluster Configuration > Layer 4 Virtual Service and click Add a New Virtual Service Enter the following details: Page 21

22 Appliance Configuration for Exchange 2013 Using DR Mode Enter an appropriate label for the VIP, e.g. CAS-IMAP4 Set the Virtual Service IP address field to the required IP address, e.g Set the Virtual Service Ports field to 143,993 Leave Protocol set to TCP Leave Forwarding Method set to Direct Routing Click Update Now click Modify next to the newly created VIP Set Balance mode to Weighted Round Robin Microsoft recommends that 'Round Robin' rather than 'Least Connection' should be used to help prevent over loading servers when they are brought online. This could occur if Least Connection was selected, since the load balancer would try to balance the number of connections across all real severs and therefore send all new requests to the new server. The trade off here is that using Round Robin will mean that server load may remain unbalanced for some time after bringing a new server into the active pool. 11. Un-check the Persistence option 12. Click Update b) Setting up the Real Servers Using the WebUI, navigate to: Cluster Configuration > Layer 4 Real Servers and click Add a new Real Server next to the newly created VIP Enter the following details: Page 22

23 Appliance Configuration for Exchange 2013 Using DR Mode Enter an appropriate label for the RIP, e.g. CAS1 Change the Real Server IP Address field to the required IP address, e.g Click Update Repeat the above steps to add your other CAS Server(s) CONFIGURE VIP3 CAS ROLE SMTP SERVICES a) Setting up the Virtual Service 2. Using the WebUI, navigate to: Cluster Configuration > Layer 4 Virtual Service and click Add a New Virtual Service Enter the following details: Enter an appropriate label for the VIP, e.g. CAS-SMTP Set the Virtual Service IP address field to the required IP address, e.g Set the Virtual Service Ports field to 25 Leave Protocol set to TCP Leave Forwarding Method set to Direct Routing Click Update Now click Modify next to the newly created VIP Un-check the Persistence option Click Update 1. b) Setting up the Real Servers Using the WebUI, navigate to: Cluster Configuration > Layer 4 Real Servers and click Add a new Real Server next to the newly created VIP Enter the following details: Page 23

24 Appliance Configuration for Exchange 2013 Using DR Mode Enter an appropriate label for the RIP, e.g. CAS1 Change the Real Server IP Address field to the required IP address, e.g Click Update Repeat the above steps to add your other CAS Server(s) 13. Appliance Configuration for Exchange 2013 Using SNAT Mode CONFIGURE LAYER 7 TIMEOUT SETTINGS To ensure that client connections remain open during periods of inactivity, the Client and Real Server Timeout values must be changed from their default values of 43 seconds and 45 seconds respectively to 30 minutes. To do this follow the steps below: 1. Using the WebUI, navigate to: Configuration > Layer 7 Advanced Configuration 2. Change Client Timeout to as shown above (i.e ms which is 30 minutes) You can also enter 30m rather than Change Real Server Timeout to as shown above (i.e ms which is 30 minutes) You can also enter 30m rather than Page 24

25 Appliance Configuration for Exchange 2013 Using SNAT Mode 4. Click the Update button to save the settings CONFIGURE VIP1 CAS ROLE HTTPS SERVICES a) Setting up the Virtual Service 2. Using the WebUI, navigate to: Cluster Configuration > Layer 7 Virtual Service and click Add a New Virtual Service Enter the following details: Enter an appropriate label for the VIP, e.g. CAS-HTTPS Set the Virtual Service IP address field to the required IP address, e.g Set the Virtual Service Ports field to 443 Set Layer 7 Protocol set to TCP Mode Click Update Now click Modify next to the newly created VIP Set Balance mode to Weighted Round Robin 1. Microsoft recommends that 'Round Robin' rather than 'Least Connection' should be used to help prevent over loading servers when they are brought online. This could occur if Least Connection was selected, since the load balancer would try to balance the number of connections across all real severs and therefore send all new requests to the new server. The trade off here is that using Round Robin will mean that server load may remain unbalanced for some time after bringing a new server into the active pool. 10. Set Health Checks to Negotiate HTTPS 11. Set Request to send to owa/healthcheck.htm As mentioned earlier, any other Exchange virtual directory (e.g. ECP, EWS etc.) can be used if preferred or more appropriate. All have an associated healthcheck.htm that can be used in the same way. Note that healthcheck.htm is generated in-memory based on the component state of the protocol in question and does not physically exist on disk. 12. Leave Response expected blank, this will configure the load balancer to look for a '200 OK' response Page 25

26 Appliance Configuration for Exchange 2013 Using SNAT Mode 13. Click Update b) Setting up the Real Servers 2. Using the WebUI, navigate to: Cluster Configuration > Layer 7 Real Servers and click Add a new Real Server next to the newly created VIP Enter the following details: Enter an appropriate label for the RIP, e.g. CAS1 Change the Real Server IP Address field to the required IP address, e.g Change the Real Server Port field to 443 Click Update Repeat the above steps to add your other CAS Server(s) 1. CONFIGURE HTTP TO HTTPS OWA REDIRECT If required, the load balancer can be configured to automatically redirect users who attempt to connect to to For details on configuring this, please refer to section 4 in the Appendix on page 37. CONFIGURE VIP2 CAS ROLE IMAP4/POP3 SERVICES a) Setting up the Virtual Service These steps show IMAP4 settings, for POP3 change the port numbers from 143 & 993 to 110 & Using the WebUI, navigate to: Cluster Configuration > Layer 7 Virtual Service and click Add a New Virtual Service Enter the following details: Page 26

27 Appliance Configuration for Exchange 2013 Using SNAT Mode Enter an appropriate label for the VIP, e.g. CAS-IMAP4 Set the Virtual Service IP address field to the required IP address, e.g Set the Virtual Service Ports field to 143,993 Set Layer 7 Protocol to TCP Mode Click Update Now click Modify next to the newly created VIP Set Balance mode to Weighted Round Robin Microsoft recommends that 'Round Robin' rather than 'Least Connection' should be used to help prevent over loading servers when they are brought online. This could occur if Least Connection was selected, since the load balancer would try to balance the number of connections across all real severs and therefore send all new requests to the new server. The trade off here is that using Round Robin will mean that server load may remain unbalanced for some time after bringing a new server into the active pool. 10. Ensure Persistence Mode is set to None 11. Click Update b) Setting up the Real Servers Using the WebUI, navigate to: Cluster Configuration > Layer 7 Real Servers and click Add a new Real Server next to the newly created VIP Enter the following details: Page 27

28 Appliance Configuration for Exchange 2013 Using SNAT Mode Enter an appropriate label for the RIP, e.g. CAS1 Change the Real Server IP Address field to the required IP address, e.g Leave the Real Server Port field blank Click Update Repeat the above steps to add your other CAS Server(s) CONFIGURE VIP3 CAS ROLE SMTP SERVICES It's important to remember that when using SNAT mode (HAProxy), the source IP address of packets reaching the Exchange Servers will be the IP address of the load balancer and not the source IP address of the client. Transparency is normally only an issue for SMTP traffic at the receive connector. System Administrators typically want to lock down receive connectors to accept SMTP connections only from a controlled set of devices such as external smart mail hosts, printers, networked photocopiers etc. If transparency for SMTP is the only issue, there are a number of options available to address this: Option 1 Enable full layer 7 transparency using TProxy. This is covered in section 1 of the Appendix on page 35. Option 2 Use the load balancers on-board firewall to lock down inbound SMTP connections rather than doing this at the receive connector. This is covered in section 2 of the Appendix on page 36. Option 3 Configure a layer 4 Virtual Service for SMTP rather than a layer 7 (HAProxy) based Virtual Service. Layer 4 is transparent by default so the source IP address is maintained. This is covered in section 3 of the Appendix on page 36. Page 28

29 Appliance Configuration for Exchange 2013 Using SNAT Mode a) Setting up the Virtual Service 2. Using the WebUI, navigate to: Cluster Configuration > Layer 7 Virtual Service and click Add a New Virtual Service Enter the following details: Enter an appropriate label for the VIP, e.g. CAS-SMTP Set the Virtual Service IP address field to the required IP address, e.g Set the Virtual Service Ports field to 25 Set Layer 7 Protocol to TCP Mode Click Update Now click Modify next to the newly created VIP Ensure Persistence Mode is set to None Click Update 1. b) Setting up the Real Servers Using the WebUI, navigate to: Cluster Configuration > Layer 7 Real Servers and click Add a new Real Server next to the newly created VIP Enter the following details: Enter an appropriate label for the RIP, e.g. CAS1 Change the Real Server IP Address field to the required IP address, e.g Change the Real Server Port field to 25 Click Update Page 29

30 Appliance Configuration for Exchange 2013 Using SNAT Mode 8. Repeat the above steps to add your other CAS Server(s) FINALIZING THE CONFIGURATION To apply the new settings, HAProxy must be restarted as follows: 1. Go to Maintenance > Restart Services and click Restart HAProxy 14. Testing & Verification USEFUL EXCHANGE 2013 & OTHER MICROSOFT TOOLS TESTING SERVER HEALTH-CHECKS USING SET-SERVERCOMPONENTSTATE The Exchange Management shell cmdlet Set-ServerComponentState can be used to verify that the load balancer is correctly health-checking the Exchange servers. In this guide, the health-check verifies that the owa virtual directory can be accessed. To verify that the health-check is working correctly, the following command can be used: Set-ServerComponentState <SERVER> -Component OwaProxy -Requester Maintenance -State Inactive Where <SERVER> is the hostname of the Exchange Server Once run, the server specified should be marked down (shown red) in the System Overview of the loadbalancer's WebUI To bring it back online, use the following command: Set-ServerComponentState <SERVER> -Component OwaProxy -Requester Maintenance -State Active Where <SERVER> is the hostname of the Exchange Server Once run, the server specified should be marked up (shown green) in the System Overview of the loadbalancer's WebUI Exchange Management Shell: TESTING MAILFLOW The Test-Mailflow cmdlet can be used to diagnose whether mail can be successfully sent and delivered. To send a test probe message to the administrators address, use the following command: Page 30

31 Testing & Verification Test-Mailflow -Target Address Exchange Management Shell: If everything is working correctly, a new message will appear in the test users mailbox: TESTING SMTP MAIL FLOW USING TELNET SMTP can be tested using telnet to connect to port 25, then by issuing various commands to simulate an being sent. Using System Overview in the WebUI, each CAS Server server can be tested by 'Halting' all others then running through the tests. To connect to port 25 of a server using Telnet, use the following command: telnet <IP Address> 25 The following screenshot shows an example of using telnet to verify SMTP operation: Page 31

32 Testing & Verification If everything is working correctly, a new message will appear in the test users mailbox: To do the same test via the load balancer, connect to the VIP rather than directly to each server, e.g.: telnet mail.robstest.com 25 MICROSOFT EXCHANGE TESTING TOOL The Remote Connectivity Analyzer tool available at is a useful Web-based Microsoft tool designed to help IT Administrators troubleshoot connectivity issues with their Exchange Server deployments. The tool simulates several client logon and mail flow scenarios. When a test fails, many of the errors have troubleshooting tips to assist the IT Administrator in correcting the problem. USEFUL APPLIANCE BASED TOOLS & FEATURES USING SYSTEM OVERVIEW The System Overview is accessed using the WebUI. It shows a graphical view of all VIPs & RIPs (i.e. the Exchange Servers) and shows the state/health of each server as well as the state of the each cluster as a whole. Th example below shows that both CAS servers are healthy and available to accept connections. Page 32

33 Testing & Verification The example below shows that rip2 has been put in halt mode: LAYER 4 STATUS REPORT The Layer 4 Status report gives a summary of layer 4 configuration and running stats as shown below. This can be accessed in the WebUI using the option: Reports > Layer 4 Status. LAYER 7 STATISTICS REPORT The Layer 7 Statistics report gives a summary of all layer 7 configuration and running stats as shown below. This can be accessed in the WebUI using the option: Reports > Layer 7 Status. Page 33

34 Testing & Verification APPLIANCE LOGS Logs are available for both layer 4 and layer 7 services and can be very useful when trying to diagnose issues. Layer 4 logs are active by default and can be accessed using the WebUI option: Logs > Layer 4. Layer 7 logging is not enabled by default (because its extremely verbose) and can be enabled using the WebUI option: Cluster Configuration > Layer 7 Advanced Configuration, and then viewed using the option: Logs > Layer Technical Support For more details about configuring the appliance and assistance with designing your deployment please don't hesitate to contact the support team using the following address: support@loadbalancer.org 16. Further Documentation The Administration Manual contains much more information about configuring and deploying the appliance. It's available here: Conclusion Loadbalancer.org appliances provide a very cost effective solution for highly available load balanced Exchange 2013 environments. Page 34

35 Appendix 18. Appendix 1 ENABLING LAYER 7 TRANSPARENCY USING TPROXY As mentioned previously, Layer 7 SNAT mode is non-transparent by default. If a fully transparent configuration is required, TProxy can be used. The main points to note are that two subnets must be used and the default gateway on the Exchange Servers must be set to be the load balancer. DAG client connections CAS/MBOX 1 VIPs LB CAS/MBOX 2 Clustered Pair Subnet 1 Subnet 2 Key points to consider: The Exchange Servers must be on a different subnet to the VIP this can be achieved by using two network adapters, or by creating VLANs on a single adapter The default gateway on the Exchange Servers must be configured to be an IP address on the load balancer. For a clustered pair of load balancers, an additional floating IP should be used for this purpose allow failover to the slave TProxy must be enabled using the WebUI: Go to Cluster Configuration > Layer 7 Advanced Configuration and set Transparent Proxy to 'On' and click Update If the load balancer has been deployed in Layer 4 DR mode, this is transparent by default so no additional steps are required. This section only applies when Layer 7 SNAT mode was initially used and transparency is now required. Page 35

36 Appendix 2 LIMITING INBOUND SMTP CONNECTIONS USING FIREWALL RULES Since layer 7 is not transparent by default, it's not possible to filter inbound SMTP connections by IP address at the receive connector on the Hub Transport Server. One way to address this is to add firewall rules to the load balancer to limit which hosts can connect inbound on port 25. Rules can be added using the WebUI option: Maintenance > Firewall Script. Simply copy/paste/edit the examples below into the firewall script then click Update. EXAMPLES: 1) to limit inbound SMTP connections to a specific smart host: VIP1=" " SRC1=" " iptables -A INPUT -p tcp --src $SRC1 --dst $VIP1 --destination-port 25 -j ACCEPT iptables -A INPUT -p tcp --dport 25 -j DROP These rules will only allow SMTP traffic from the host to reach the VIP. 2) to limit inbound SMTP connections to a range of smart hosts: VIP1=" " SRC1=" " iptables -A INPUT -p tcp -m iprange --src-range $SRC1 --destination $VIP1 --destination-port 25 -j ACCEPT iptables -A INPUT -p tcp --dport 25 -j DROP These rules will only allow SMTP traffic from hosts in the range through to reach the VIP. If the load balancer has been deployed in Layer 4 DR mode, this is transparent by default so no additional steps are required. This section only applies when Layer 7 SNAT mode was initially used and transparency is now required. 3 USING A LAYER 4 VIRTUAL SERVICE FOR SMTP Layer 7 Virtual Services are not transparent by default which can be an issue for the HT role. One option In this case is to use a Layer 4 DR mode VIP. For more details about Layer 4 DR mode please refer to page 9. If the load balancer has been deployed in Layer 4 DR mode, this is transparent by default so no additional steps are required. This section only applies when Layer 7 SNAT mode was initially used and transparency is now required. Page 36

37 Appendix Layer 4 DR Mode - Solving the ARP Problem: Layer 4 DR mode works by changing the MAC address of the inbound packets to match the Real Server selected by the load balancing algorithm. To enable DR mode to operate: Each Real Server must be configured to accept packets destined for both the VIP address and the Real Servers IP address (RIP). This is because in DR mode the destination address of load balanced packets is the VIP address, whilst for other traffic such as health-checks, administration traffic etc. it's the Real Server's own IP address (the RIP). The service/process (e.g. IIS) must respond to both addresses. Each Real Server must be configured so that it does not respond to ARP requests for the VIP address only the load balancer should do this. Configuring the Real Servers in this way is referred to as 'Solving the ARP problem''. The steps required depend on the particular version of Windows being used. For detailed steps on solving the ARP problem for Windows 2012/2016 Please refer to page CONFIGURING AN HTTP TO HTTPS REDIRECT FOR OWA An additional layer 7 VIP is required that listens on HTTP port 80 on the same IP address. The VIP is then configured to redirect connections to HTTPS port 443. e.g. should be redirected to The steps: 1) Create another Layer 7 VIP with the following settings: Label: HTTP-redirect Virtual Service IP Address: <same as the VIP that's listening on port 443> Virtual Service Ports: 80 Layer 7 Protocol: HTTP Mode Persistence Mode: None Force to HTTPS: Yes This additional VIP will be shown purple/green to indicate that it's being used for HTTP to HTTPS redirection. 2) Apply the new settings to apply the new settings, HAProxy must be restarted: Using the WebUI, navigate to: Maintenance > Restart Services and click Restart HAProxy Page 37

38 Appendix 5 CLUSTERED PAIR CONFIGURATION ADDING A SLAVE UNIT If you initially configured just the master unit and now need to add a slave - our recommended procedure, please refer to the relevant section below for more details: A number of settings are not replicated as part of the master/slave pairing process and therefore must be manually configured on the slave appliance. These are listed below: Hostname & DNS settings SNMP settings Network settings including IP addresses, bonding configuration and VLANs Routing configuration including default gateways and static routes Date & time settings Physical Advanced Configuration settings including Internet Proxy IP address & port, Firewall table size, SMTP relay and Syslog server Graphing settings Firewall Script & Firewall Lockdown Script settings Software updates Version 7: Please refer to Chapter 8 Appliance Clustering for HA in the v7 Administration Manual. Version 8: To add a slave node i.e. create a highly available clustered pair: Deploy a second appliance that will be the slave and configure initial network settings Using the WebUI, navigate to: Cluster Configuration > High-Availability Configuration Page 38

39 Appendix Specify the IP address and the loadbalancer users password (the default is 'loadbalancer') for the slave (peer) appliance as shown above Click Add new node Once complete, the following will be displayed: To finalize the configuration, restart heartbeat and any other services as prompted in the blue message box at the top of the screen The pairing process now commences as shown below: Clicking the Restart Heartbeat button on the master appliance will also automatically restart heartbeat on the slave appliance. Please refer to chapter 9 Appliance Clustering for HA in the Administration Manual for more detailed information on configuring HA with 2 appliances. Page 39

40 Appendix 6 SOLVING THE ARP PROBLEM When using Layer 4 DR mode, the ARP problem must be solved. This involves configuring each Real Server to be able to receive traffic destined for the VIP, and ensuring that each Real Server does not respond to ARP requests for the VIP address only the load balancer should do this. The steps below are for Windows 2012 / 2016, for other versions of Windows please refer to chapter 6 in the Administration Manual. Step 1: Install the Microsoft Loopback Adapter Click Start, then run hdwwiz to start the Hardware Installation Wizard When the Wizard has started, click Next Select Install the hardware that I manually select from a list (Advanced), click Next Select Network adapters, click Next Select Microsoft & Microsoft KM-Test Loopback Adapter, click Next 6. Click Next to start the installation, when complete click Finish Step 2: Configure the Loopback Adapter Open Control Panel and click Network and Sharing Center Click Change adapter settings Right-click the new Loopback Adapter and select Properties Un-check all items except Internet Protocol Version 4 (TCP/IPv4) and Internet Protocol Version 6 (TCP/IPv6) as shown below: Page 40

41 Appendix Leaving both checked ensures that both IPv4 and IPv6 are supported. Select one If preferred. 5. If configuring IPv4 addresses select Internet Protocol Version (TCP/IPv4), click Properties and configure the IP address to be the same as the Virtual Service (VIP) with a subnet mask of , e.g / as shown below: Page 41

42 Appendix 6. If configuring IPv6 addresses select Internet Protocol Version (TCP/IPv6), click Properties and configure the IP address to be the same as the Virtual Service (VIP) and set the Subnet Prefix Length to be the same as your network setting, e.g. 2001:470:1f09:e72::15/64 as shown below: 7. Click OK on TCP/IP Properties, then click Close on Ethernet Properties to save and apply the new settings 8. Now repeat the above process on the other Windows 2012/2016 Real Servers Step 3: Configure the strong/weak host behavior Windows Server 2000 and Windows Server 2003 use the weak host model for sending and receiving for all IPv4 interfaces and the strong host model for sending and receiving for all IPv6 interfaces. You cannot configure this behavior. The Next Generation TCP/IP stack in Windows 2008 and later supports strong host sends and receives for both IPv4 and IPv6 by default. To ensure that Windows 2012/2016 is running in the correct mode to be able to respond to the VIP, the following commands must be run on each Real Server: For IPv4 addresses: netsh interface ipv4 set interface "net" weakhostreceive=enabled netsh interface ipv4 set interface "loopback" weakhostreceive=enabled netsh interface ipv4 set interface "loopback" weakhostsend=enabled For these commands to work, the LAN connection NIC must be named net and the loopback NIC must be named loopback as shown below. If you prefer to leave your current NIC names, then the commands above must be modified accordingly. For example, if your network adapters are named LAN and LOOPBACK, the commands required would be: netsh interface ipv4 set interface "LAN" weakhostreceive=enabled netsh interface ipv4 set interface "LOOPBACK" weakhostreceive=enabled netsh interface ipv4 set interface "LOOPBACK" weakhostsend=enabled For IPv6 addresses: Page 42

43 Appendix netsh interface ipv6 set interface "net" weakhostreceive=enabled netsh interface ipv6 set interface "loopback" weakhostreceive=enabled netsh interface ipv6 set interface "loopback" weakhostsend=enabled netsh interface ipv6 set interface "loopback" dadtransmits=0 For these commands to work, the LAN connection NIC must be named net and the loopback NIC must be named loopback as shown below. If you prefer to leave your current NIC names, then the commands above must be modified accordingly. For example, if your network adapters are named LAN and LOOPBACK, the commands required would be: netsh interface ipv6 set interface "LAN" weakhostreceive=enabled netsh interface ipv6 set interface "LOOPBACK" weakhostreceive=enabled netsh interface ipv6 set interface "LOOPBACK" weakhostsend=enabled netsh interface ipv6 set interface "LOOPBACK" dadtransmits=0 The names for the NICs are case sensitive, so make sure that the name used for the interface and the name used in the commands match exactly. 1. Start Powershell or use a command window to run the appropriate netsh commands as shown in the example below: This shows an IPv6 example, use the IPv4 commands if you're using IPv4 addresses. 2. Now repeat these 4 commands on the other Windows 2012 Real Servers Solving the ARP problem for other version of Windows is similar. For full details, please refer to the Administration Manual. Page 43